rediris.cf

harvest是一个下载html网页得机器人

# $Id: modern.cf,v 1.1 2002/07/05 by Harald Weinreich <[e-mail]>
#
# Definitions for '{nph-}search.cgi' configuration.
#
# A hash (#) in the first column denotes a comment and is not processed
# even inside a definition.  Hashes not in the first column are left as-is.
#
# Variable substitution occurs on these definitions.  If you want
# a dollar-sign ($) to occur in the output, escape it with backslash (\).
# Other metacharacters (quotes, asterisks, etc) probably don't need to be
# escaped.  Printf-like special characters are allowed: \n \t \r etc.
#
# The ending newline is chopped off from each definiton.  So
#
#   <FooBar>
#   abcxyz
#   </FooBar>
#
# becomes "abcxyz".  You may have blank lines in between the beginning
# and ending tags, or use the newline character \n.
#

#
# CGI defaults.
#
# In this section you can give default values for any attribute that would
# normally be passed in from the CGI query form.  The following attributes
# are currently available:
#
# brokerqueryconfig : Name of a broker specific configuration file
# lifetime          : Maximum lifetime of a query (see also <Lifetime> section)
# caseflag          : Whether the query is casesensitive ["on" or "off"]
# wordflag          : Match on word boundaries ["on" or "off"]
# opaqflag          : Show matched lines ["on" or "off"]
# descflag          : Return object description ["on" or "off"]
# noregexflag       : Don't do regular expressions ["on" or "off"]
# maxresultflag     : Maximum number of results to return
# maxobjflag        : Maximum number of objects to return
# maxlineflag       : Maximum number of matched lines
# errors            : Number of spelling errors
# broker            : Name of broker
# host              : Name of broker host
# attributes        : Space sperated list of attributes to return
# sort              : Sort options ["by-rank"]

# These defaults mimic the values on the provided query-glimpse-modern.html form.
<Default>
caseflag      : on
wordflag      : on
errorflag     : 0
opaqueflag    : on
descflag      : on
maxobjflag    : 500
maxlineflag   : 30
maxresultflag : 3000
perpageflag   : 25
sort          : by-rank
</Default>

# 'GLOBAL' VARIABLES: are defined globally within the {nph-}search.cgi program,
#                     but not necessarily set at all times.
#
# $query	the user query string
# $html_query	the user query string, special HTML characters escaped
# $bquery	the query string sent to the broker
# $host		the broker hostname
# $port		the broker port
# $hp_url	the URL of the broker query (home) page
# $maxresult	the maximum number of matched lines the broker will return
# $nobjects	a running count of the number of objects returned
# $nopaquelines a running count of the number of opaque (matched) lines
#
# definitions here can be output in other definitions.  Each definition
# here is placed into the %CFG associative array.  For example,
# to print out the Timeout defined below, write $CFG{'Timeout'}.
#

# The URL to the Harvest Project home page.
#
<HarvestUrl>
http://harvest.sourceforge.net/
</HarvestUrl>

# The amount of time to wait for the broker results.
# Can use aritmetic # here also, eg: 5*60
#
# We use a conservative 5 minutes here.
#
<Timeout>
300
</Timeout>

# InitFunction can be some perl code which gets eval'd before they
# query is sent.  Use this for any special hackings.
#
<InitFunction>
$cs_urlX = $csumflag ? 'Y' : undef;
</InitFunction>

# The MIME content type of the query results.  Should be "text/html"
# if we are going to return HTML tags.
#
<ContentType>
text/html
</ContentType>


##############################################################################
# RESULT SECTION
#
# The output of '{nph-}search.cgi' consists of the following tags:
#
#    ResultHeader
#    CreateNavBars
#    ResultSetBegin
#    ( Errors from 'broker' )
#    foreach object {
#        PrintObject
#    }
#    ResultSetEnd
#    EndBrokerResults
#    ( TruncateWarning )
#    ResultTrailer
#
#
##############################################################################

# First output for the HTML containing the results.  Should probably include
# <TITLE> tag and the user query.
#
<ResultHeader>
<HEAD>
<TITLE>Search Results for: $html_query</TITLE>
<meta http-equiv="content-type" content="text/html; charset=ISO-8859-1">
<style><!--
.rNavBar {
  color: #6699cc;
  font-size: small;
  font-family: arial, helvetica, sans-serif;
}
.rMoreInfo {
  color: #6699cc;
  font-size: small;
  font-family: arial, helvetica, sans-serif;
  font-style: italic;
  text-decoration: none;
}
dt { margin-left:1px; }
dd { margin-left:40px; }
// --></style>
</HEAD>
<BODY bgcolor=white>
<H2>Search Results for: <font color="#800000">$html_query</font></H2>
<HR>
</ResultHeader>

# Final HTML section for the output.  Should probably contain a pointer
# to the broker home-page ($hp_url) and a pointer to the Harvest
# home page.
#
<ResultTrailer>
#----------------------------------------------------------------------
# JMM - 20011020
# We close the HTML page
#---------------------------------------------------------------------,
<table border="0" cellpadding="0" cellspacing="0" width="100%">
<tr align="center">
  <td width="100%">
    $rNavigationBar
  </td>
  <td><a href="$CFG{'HarvestUrl'}"><img
         alt="Powered by harvest" border="0" src="$rHarvestIcon"></a></td>
</tr>
</table>
</body>
</html>
#---------------------------------------------------------------------'
</ResultTrailer>


#----------------------------------------------------------------------
# JMM 20011020
# Calculations to obtain $rResultsBar and $rNavigationBar
# We use them in <ResultSetBegin>
#---------------------------------------------------------------------,
<CreateNavBars>

if ($rTotHtmlFiles > 0)
{
  $rCurrentPage = (!$current_page) ? 1 : $current_page;

  $rPrevPage = "";
  if ($rCurrentPage > 1)
  {
    $rPrevPage = $rCurrentPage - 1 ;
    $rPrevPage = "\n\n<a class=\"rNavBar\" href=\"$tmp_www/$^T-${$}-$rPrevPage.html\">&lt;&lt; Prev</a>